Single-Molecule Imaging of EGFR Signalling on the Surface of Living Cells

articles Single-molecule imaging of EGFR signalling on the surface of living cells

Yasushi Sako*, Shigeru Minoguchi* and Toshio Yanagida*†‡ *Department of Physiology and Biosignalling, Graduate School of Medicine, Osaka University, 2-2 Yamadaoka, Suita 565-0871, Japan †Single Molecule Processes Project, ICORP, JST, 2-4-14 Senba-Higashi, Mino 562-0035, Japan ‡e-mail: [email protected]

The early events in signal transduction from the epidermal growth factor (EGF) receptor (EGFR) are dimerization and autophosphorylation of the receptor, induced by binding of EGF. Here we observe these events in living cells by visualizing single molecules of fluorescent-dye-labelled EGF in the plasma membrane of A431 carcinoma cells. Single- molecule tracking reveals that the predominant mechanism of dimerization involves the formation of a cell-surface complex of one EGF molecule and an EGFR dimer, followed by the direct arrest of a second EGF molecule, indicating that the EGFR dimers were probably preformed before the binding of the second EGF molecule. Single-molecule fluorescence-resonance energy transfer shows that EGF–EGFR complexes indeed form dimers at the molecular level. Use of a monoclonal antibody specific to the phosphorylated (activated) EGFR reveals that the EGFR becomes phosphorylated after dimerization.

ingle-molecule imaging is an ideal technology by which to Results study the molecular mechanisms of biological reactions in Visualization of single molecules on the surface of living cells. As Svitro1–3. Here we extend this technology to real-time obser- a probe for detection of single molecules in living cells, we used vations of the binding of fluorescent-dye-labelled EGF to its a fluorescent dye, Cy3, conjugated to mouse EGF. As mouse EGF receptor (EGFR), dimerization of EGF–EGFR complexes, and has only one reactive amino residue (at the amino terminus), it autophosphorylation of EGFR in the plasma membrane. can be labelled with amino-reactive Cy3 dye with a dye:protein Although dimerization and autophosphorylation are known to ratio of exactly 1:1. be essential for signal transduction of EGFR4,5, the mechanisms Under an objective-type total internal reflection (TIR) fluores- underlying these events have not been fully understood until cence microcope3, Cy3-labelled EGF (Cy3–EGF) was added to the now. culture medium of A431 cells (Fig. 1). The background autofluores-

a Evanescent field c Focus Apical Basal Medium

Coverslip Oil θ

Objective TIR (apical)

Laser 1 µm

b Apical imaging 10 µm

Basal imaging TIR (basal)

Illumination

θ1

Laser θ2 θ θ Epi 1> 2

Figure 1 TIR microscopy for visualization of single molecules on the cell on both apical and basal surfaces. In this case, spots on the basal surface would be surface. a, TIR microscopy. An objective-type TIR microscope was used. b, For excited by epifluoresence illumination. Because of excitation of Cy3–EGF in the imaging of the basal cell surface, a laser beam was reflected between a coverslip basolateral surface and free Cy3–EGF in the evanescent field, background and the culture medium. For imaging of the apical plasma membrane, the incident fluorescence was higher on the apical surface than on the basal surface. Even in this θ θ ) angle of the laser beam was reduced ( 1 > 2 and the laser beam was reflected condition, individual Cy3–EGF molecules could be visualized on the apical surface between the cytoplasm and culture medium. c, To demonstrate the effect of the (inset). Under basal TIR mode, spots were only observed on the basal surface illumination mode, images of A431 cells were taken in the same field under different because of the limited excitation depth of TIR. Under the epifluorescence illumination modes in the presence of 100 ng ml–1 Cy3–EGF. The cells were 2–3 µm configuration, background from fluorescent EGF in the culture medium was high and thick, and so, by changing the focus, we could easily distinguish between the apical no fluorescent spots on the cell surface were detected. surface and the basal surface. Under apical TIR mode, fluorescent spots were shown

acb

d Cy3-EGF e 800 25 400 20 100µW µm Ð2 0 15 800 τ=2.5s 10 400 5 0 0 800 0 4 8 1216 20 400 15 0 10 τ=6.8s 40µW µm Ð2 800 5 Frequency (%) Frequency 400 0 0 0 4 8 1216 20 15

Fluorescence intensity (AU) Cy5-EGF 800 10 τ=13.9s 20µW µmÐ2 400 5 0 0 0 246810 0 4 8 1216 20 Time (s) Duration (s)

Figure 2 Visualization of Cy3–EGF on the plasma membrane of living A431 (30 frames every second) and plotted against time. An HeNe laser (632.8 nm) was cells. a, b, Cy3–EGF at a final concentration of 0.2 ng ml–1 was added to the used for observation of Cy5–EGF. AU, arbitrary unit. e, Histograms of the time culture medium of living A431 cells under a TIR fluorescence microscope. Images difference between the appearance and disappearance of Cy3–EGF spots under were acquired immediately (a) and 1 min (b) after the addition of Cy3–EGF. c, A lasers of variable excitation powers. The lines represent the fitting of the results mixture of Cy3–EGF (0.2 ng ml–1) and EGF (200 ng ml–1) was added to the culture to a single exponential function. τ indicates the time constant of the fit function. medium of A431 cells. This image was acquired 1 min after the addition of Cy3– Multiple correlation coefficients (R) of the fitting are 0.991 (100 µW µm–2), 0.916 EGF and EGF. Scale bar represents 10 µm. d, The change in fluorescence intensity (40 µW µm–2) and 0.741 (20 µW µm–2). of individual Cy3 (or Cy5)–EGF spots on living cells was measured frame by frame cence of cells before addition of Cy3–EGF was low, except at large lifetime is proportional to the excitation laser power (Fig. 2e). This vesicles in the perinuclear region. Soon after the addition of Cy3– relationship between the lifetime and excitation power indicates EGF, fluorescent spots appeared on the cell surface; the density of that the duration of fluorescence was determined by the rate of pho- these spots increased with time (Fig. 2a, b). Cy3–EGF was not tobleaching of Cy3–EGF. Thus it is likely that the sudden disappear- detectable as fluorescent spots in the culture medium because of ance of the Cy3–EGF spots was caused by photobleaching, not by brownian movement in solution. In addition, the background flu- dissociation from EGFR or by endocytosis into the cytoplasm. Sin- orescence caused by Cy3–EGF in the medium was low because of gle-step photobleaching is strong evidence that the fluorescent the low excitation depth of TIR microscopy. Binding of Cy3–EGF spots represented single Cy3–EGF molecules on the cell surface1. to the cell surface was completely inhibited by addition of excess We conjugated a monoclonal anti-EGFR antibody (EGFR1) unlabelled EGF at the same time that Cy3–EGF was added (Fig. 2c), with Cy3 (generating Cy3–EGFR1), and allowed this to bind to indicating specific binding of Cy3–EGF to the cell-surface EGFR. A431 cells in the same way as Cy3–EGF. The concentration of Cy3– Cy3–EGF did not bind to CHO-K1 cells, which lack EGFR6 (data EGFR1 was low (0.1 µg ml–1) so as not to induce aggregation of not shown). EGFR. As EGFR1 can be labelled with multiple Cy3 molecules, the By carefully adjusting the incident angle of the excitation laser intensity distribution of Cy3–EGFR1 spots on the cell surface was beam, we were able to visualize Cy3–EGF on both apical and baso- fitted to a sum of two or three Gaussian distribution functions (Fig. lateral plasma membranes by TIR microscopy (Fig. 1b, c). A slight 3b, c). The results of this fitting are consistent with Poisson distri- difference in the refractive indices of the cytoplasm and the culture butions of the number of Cy3 molecules to EGFR1 molecules, indi- medium resulted in imaging of the apical surface of the cell mem- cating that the first, the second and the third components represent brane. Here we studied mainly the apical surface (see Supplemen- EGFR1 conjugated with one, two or three Cy3 molecules, respec- tary Information). tively. We followed the change in fluorescence intensity of each Cy3– As, in the intensity distribution of Cy3–EGF (Fig. 3a), the mean EGF spot in living A431 cells (Fig. 2d). In many cases, a spot of the first component was the same as those for Cy3–EGFR1, we appeared suddenly on the cell surface, emitted almost constant flu- conclude that this component contains single Cy3–EGF molecules. orescence, then disappeared suddenly. Histograms of the time dif- Binding of EGF to EGFR induces dimerization of EGFR4, which is ference between the appearance and the disappearance of each spot indispensable for EGF signalling. The second component, which were well fit with a single exponential function. The inverse of the emits a signal twice as intense as the first component, may represent

86 a + Cy3-EGF a 16 4 3 Cy3-EGF 12 2 n=195 intensity 1 8 012 34 14 Relative fluorescence Relative Time (min) 4 b 16 0 0 500 1,000 1,500 54 0-20s b 16 12 83 (81) 46 n=141 Cy3-EGFR1 8 12 D/P=0.4 4 8 n=200 0 17 (16) 0 100 200 300 400 500 4

Frequency (%) Frequency 16 37 0 63 0 500 1,000 1,500 12 20-40s c 16 n=250 63 (60) 8 Cy3-EGFR1 12 D/P=1.0 4 n=192 8 (%) Frequency 0 35 (30) 0 100 200 300 400 500 4 2 (10) 16 31 0 12 40-60s 0 500 1,000 1,500 69 n=234 Fluorescence intensity (AU) 8

Figure 3 Distribution of the fluorescence intensity of Cy3–EGF and Cy3– 4 EGFR1 1 on the surface of living cells. As well as Cy3–EGF (a), two 0 preparations of Cy3-labelled anti-EGFR antibody (Cy3–EGFR1; final concentration of 0 100 200 300 400 500 0.1 µg ml–1; b, c) with different average dye:protein ratios (D/P) were applied to A431 Fluorescence intensity (AU) cells. Histograms of the fluorescence intensity were fit to a sum of two (a, b) or three 2+ (c) Gaussian distributions (lines). The peak, mean and standard deviation of Gaussian Figure 4 Intracellular Ca response induced by Cy3–EGF and the change in functions were fit. No parameter was forced on another. The positions of the mean intensity distribution of Cy3–EGF spots. a, Cells were loaded with the 2+ –1 (arrows) and percentage of the fraction (numbers not in parentheses) of each fluorescent Ca indicator Fluo-3 (Dojindo, Kumamoto, Japan). At time 0, 1 ng ml Gaussian component are indicated. Numbers in parentheses are the percentage of Cy3–EGF was added to the culture medium at 25 °C. The increase in the the fractions expected from Poisson distributions with average of 0.4 (b) and 1.0 (c). fluorescence intensity of Fluo-3 over time indicates an increase in the concentration 2+ Multiple correlation coefficients (R) of the fitting are 0.967 (b) and 0.977 (c). of intracellular Ca . b, Fluorescence-intensity distributions of Cy3–EGF spots were studied at various times after the addition of Cy3–EGF. The experimental conditions were the same as those for a. The fluorescence intensity of the spots was measured two Cy3–EGF molecules bound to a dimer of EGFR or two EGFR in 11–15 single-frame snapshots of different cells during each period. Results were monomers closer than the spatial resolution of the microscope. The fitted to a sum of two Gaussian distributions (line), and the percentages of fractions latter proposal is less probable because of the low density of the flu- of each component are indicated (numbers). The peak, mean and standard deviation orescent spots. of Gaussian functions were fit. No parameter was forced on another. Multiple From the results shown in Figs 2, 3, we conclude that we had vis- correlation coefficients (R) of the fitting are 0.945 (0–20 s), 0.933 (20–40 s) and ualized single Cy3–EGF molecules on the surface of living cells. We 0.954 (40–60 s). The distributions at 20–40 s and 40–60 s are statistically similar to visualized single molecules of EGF labelled with other fluorescent the distribution at 0–20 s, with a significance (P) of 0.38 and 0.0065, respectively dyes, Cy5 (Fig. 2d) and tetramethylrhodamine (data not shown), by (Mann–Whitney U-test). Densities of the Cy3–EGF spots on the cell surface were the same technique. 0.18, 0.38 and 0.42 per µm2 at 0–20, 20–40 and 40–60 s, respectively. Single-molecule tracking of EGFR dimerization. The intracellular Ca2+ response induced by EGF requires dimerization of the receptor7. Thus, dimerization should take place before the increase of fluorescent spots containing two Cy3–EGF molecules. In a few in the intracellular Ca2+ concentration that we observed 1 min after cases (3 out of 26), two spots diffusing laterally along the plasma the addition of Cy3–EGF (Fig. 4a). We studied the formation of the membrane collided and then moved together (Fig. 5a, upper EGF–EGFR dimeric complexes by checking the intensity distribu- panel). However, in most cases (23 out of 26), the fluorescence tion of the fluorescent spots at various times after the addition of intensity of a spot increased suddenly by a factor of about two, Cy3–EGF. The amount of the component containing two Cy3–EGF then decreased in two steps (Fig. 5a, lower panel, Fig. 5b). This molecules increased 40–60 s after the addition of Cy3–EGF (Fig. result points to a dimerization between EGFR molecules that 4b), as the density of cell-surface Cy3–EGF spots increased (Fig. 2a, exists just before the binding of a second EGF molecule. It is b). This increase in levels of the fraction containing EGFR dimer probable that the preformed dimer, which could not be detected before the increase in Ca2+ amounts was consistent with the theory using previous methods5,10–12, is formed before the second EGF that signal transduction from the EGFR occurs through EGFR molecule binds. dimerization. Usually, lateral mobility of Cy3–EGF spots on the cell surface Single-molecule visualization enables single-molecule track- was small; large-scale movement is not necessary for EGFR sig- ing in living cells8,9. Using this technique, we traced the formation nalling. One reason for the restriction of lateral mobility of EGFR

a Time (s) a Cy3-EGF Cy5-EGF Merged 0:00 0:20 0:40 0:60 1:002:50 3:50 4:00

0:00 1:00 2:00 3:00 4:00 5:00 6:00 7:00

Cy3-EGF b 1,000 750 Cy5-EGF 500 250 Merged

Fluorescence intensity (AU) 0 Ð2 02468 Time (s) 0:00 1:00 2:00

1 Figure 5 Dimerization of EGF–EGFR complexes. a, Under the same conditions Cy3 Cy5 as those used for Fig. 3a, the formation of Cy3–EGF spots with a signal intensity 0.5 twice that of monomer spots was traced. In the upper panel, two spots (arrow and arrowhead) collided at time 0:40 s and then moved together. The fluorescence Normalized 0 intensity increased after the collision. At 4:00 s, the intensity of the spot decreased

fluorescence intensity 0:00 1:00 2:00 to about half that at 3:50 s (arrow), probably because of photobleaching of one Cy3– Time (s) EGF molecule. In the lower panel, a fluorescence spot with the intensity of a single molecule was observed until 1:00 s; intensity of the spot increased suddenly Figure 6 Single-molecule FRET between Cy3–EGF and Cy5–EGF. a, A mixture between 1:00 s and 2:00 s, then decreased between 4:00 s and 5:00 s. Scale bar of Cy3–EGF (0.2 ng ml–1) and Cy5–EGF (0.8 ng ml–1) was added to the culture medium represents 5 µm. b, Time trace of the fluorescence intensity of the spot shown in the of A431 cells. The specimen was excited with a green (532 nm) laser beam and lower panel of a. A two-step build-up and bleaching of intensity can be seen. The time fluorescence images of the same field in the Cy3 channel (565–595 nm) and Cy5 trace is noisy because of slow, lateral diffusive movement of the spot. channel (650–690 nm) were acquired simultaneously by using dual-view optics. Fluorescent spots shown in the Cy5 channel appeared as a result of FRET from Cy3– EGF to Cy5–EGF (arrows). b, Top three rows, the change in fluorescence intensity is binding to the cytoskeleton13,14. over time of a single spot (arrow) was measured in both Cy3 and Cy5 channels. The position of the spot moved slowly, probably because of slow brownian movement of Single-molecule fluorescence-resonance energy-transfer analysis. 25 We measured molecular-level interactions of EGF in an EGFR cytoskeleton-bound membrane proteins . To avoid photobleaching, the excitation µ µ –2 dimer by single-molecule fluorescence-resonance energy transfer laser power was reduced to 10 W m and images were acquired using a four- (FRET)2,15. We added a mixture of Cy3–EGF and Cy5–EGF to the frame rolling average. Bottom, the fluorescence intensities in Cy3 and Cy5 channels cells, excited Cy3–EGF with a 532-nm laser, and acquired fluores- were plotted against time after normalization with respect to the peak intensity of cence images at 580 nm (for Cy3) and 670 nm (for Cy5) simultane- each channel. Anti-correlation between Cy3 and Cy5 fluorescence intensities ously by using dual-view optics16 (Fig. 6a). Anti-correlation of the indicates FRET from Cy3 to Cy5. fluorescence-intensity changes of the same spot in the Cy3 channel and the Cy5 channel indicated FRET from Cy3–EGF to Cy5–EGF, Cy5-EGF Cy3-mAb74 that is, single-molecular detection of interaction between EGF molecules in a dimer (Fig. 6b). Fluctuation of the FRET efficiency may be caused by conformational fluctuation of the dimer. The low mobility of the dimer shown in Fig. 6b was compatible with the fact that micrometre-scale lateral mobility is not necessary for dimerization and signalling of the EGFR. The probability of detecting FRET (10/201 = 5%) was lower than that expected from the histograms shown in Fig. 4. The distance between Cy3 and Cy 5 in a dimeric EGF–EGFR complex may be longer than the Förster distance between a Cy3–Cy5 pair (5.8 nm)15 (see Supplementary Information). Detection of autophosphorylation of EGFR. Dimerization of the EGFR induces autophosphorylation on its cytoplasmic tyrosine Figure 7 Phosphorylation of EGF–EGFR complexes. A431 cells were perforated residues5. A monoclonal antibody, mAb74, recognizes the confor- using streptolysin O and stimulated with 10 ng ml–1 Cy5–EGF for 1 min. Phosphorylation mational change of the cytoplasmic domain of EGFR that is of the EGFR was detected with Cy3–mAb74 (mAb74 is an antibody that recognizes induced by autophosphorylation17. We perforated A431 cells with activated (phosphorylated) EGFR). Images were acquired 15 min after the addition of streptolysin O18 to introduce Cy3-labelled mAb74 (Cy3–mAb74) Cy3–mAb74 using dual-view optics. Cy3–mAb74 bound at the positions indicated by into the cytoplasm. After stimulating the cells with Cy5–EGF, we the bright Cy5–EGF spots (arrows). Scale bar represents 10 µm. visualized the binding of Cy3–mAb74 to the plasma membrane. Fluorescent spots of Cy3–mAb74 co-localized with the spots of Cy5–EGF that had a signal about twice as intense of that of other Discussion Cy5–EGF spots, indicating autophosphorylation of EGFR after Here, using single-molecule imaging techniques, we have been dimerization (Fig. 7). This result is consistent with a widely able to detect the location, movement, interaction and biochem- accepted hypothesis for the mechanism of EGFR activation, that is, ical reaction of single EGFR molecules. Our results indicate the that formation of a dimeric EGF–EGFR complex induces autophos- presence of a preformed EGFR dimer and conformational fluc- phorylation of EGFR. tuation of this dimer.

Despite the fact that dimerization of the EGFR is known to be optics. No crosstalk between the Cy3 and Cy5 channels was observed under this optical set-up. essential for signal transduction, the process of dimerization has RECEIVED 16 SEPTEMBER 1999; REVISED 29 DECEMBER 1999; ACCEPTED 21 JANUARY 2000; PUBLISHED 10 FEBRUARY 2000. not been fully understood. Our results indicate that EGF–(EGFR)2 complexes are formed before the second EGF molecule binds. It has 1. Funatsu, T., Harada, Y., Tokunaga, M., Saito, K. & Yanagida, T. Imaging of single fluorescent molecules and individual ATP turnovers by single myosin molecules in aqueous solution. Nature 374, been reported that the binding of EGF to an EGFR dimer is more 555–559 (1995). 19– than ten times stronger than EGF binding to an EGFR monomer 2. Weiss, S. Fluorescence spectroscopy of single biomolecules. Science 283, 1676–1683 (1999). 21 . This is probably the reason why the second EGF molecule bound 3. Tokunaga, M., Kitamura, K., Saito, K., Hikikoshi-Iwane, A. & Yanagida, T. Single molecule imaging of fluorophores and enzymatic reactions achieved by objective-type total internal reflection effectively bound to EGF–(EGFR)2 in our experiments, even though there might be many unoccupied EGFR monomers on the cell sur- fluorescence microscopy. Biochem. Biophys. Res. Commun. 235, 47–53 (1997). 4. Yarden, Y. & Schlessinger, J. Epidermal growth factor induces rapid, reversible aggregation of the face. It has been suggested that hetrodimerization between EGFR- purified epidermal growth factor receptor. Biochemistry 26, 1443–1451 (1987). related receptors (ErbB2 and ErbB3) is induced by bivalency of a 5. Yarden, Y. & Schlessinger, J. Self-phosphorylation of epidermal growth factor receptor: evidence for ligand, the differentiation factor Neu22. Dimerization of the EGFR a model of intermolecular allosteric activation. Biochemistry 26, 1434–1442 (1987). may similarly be induced by bivalent binding of EGF. A twisted 6. Chung, C., Sciaky, N. & Gross, D. J. Heterogeneity of epidermal growth factor binding kinetics on model has been proposed for dimerization of the EGFR12. In this individual cells. Biophys. J. 73, 1089–1102 (1997). 7. Carraway, K. L. III & Cerione, R. A. Comparison of epidermal growth factor (EGF) receptor-receptor model, binding of EGF activates the EGFR by inducing a twisting of interactions in intact A431 cells and isolated plasma membrane. J. Biol. Chem. 266, 8899–8906 a preformed EGFR dimer. Our results are consistent with this (1991). model. Another possible interpretation of our results is that dimer- 8. Schütz, G. J., Schindler, H. & Schmidt, T. H. Single-molecule microscopy on model membranes ization occurs between EGFR molecules corralled into the dense reveals anomalous diffusion. Biophys. J. 73, 1073–1080 (1997). 23–25 9. Ide, T. & Yanagida, T. An artificial lipid bilayer formed on an agarose-coated glass for simultaneous membrane cytoskeletal network . electrical and optical measurement of single ion-channels. Biochem. Biophys. Res. Commun. 265, 595– On the cell surface, EGFR molecules with high (Kd < 1 nM) and 599 (1999). low (Kd > 10 nM) affinity for EGF are present. At the concentration 10. Azevedo, J. R. & Johnson, D. A. Temperature-dependent lateral and transverse distribution of the of Cy3–EGF used in our experiments (0.2 nM), high-affinity EGFR, epidermal growth factor receptor in A431 plasma membranes. J. Memb. Biol. 118, 215–224 (1990). which is responsible for signal transduction13, was mainly visual- 11. Zidovetzki, R., Yarden, Y., Schlessinger, J. & Jovin, T. M. Microaggregation of hormone-occupied epidermal growth factor receptors on plasma membrane preparations. EMBO J. 5, 247–250 (1986). ized. The activation of low-affinity receptors may occur in different 12. Gadella, T. W. J. Jr & Jovin, T. M. Oligomerization of epidermal growth factor receptors on A 431 ways. The conformational fluctuation of EGFR dimers is also sug- cells studied by time-resolved fluorescence imaging microscopy. A stereochemical model for tyrosine gested by the results of our single-molecule experiments. It is possi- kinase receptor activation. J. Cell Biol. 129, 1543–1558 (1995). ble that the EGFR dimer can exist in several different 13. den Hartigh, J. C., van Bergen en Henegouwen, P. M., Verkleij, A. J. & Boonstra, J. The EGF receptor conformations. Future work will address the relationship between is an actin-binding protein. J. Cell Biol. 119, 349–355 (1992). 14. Wiegant, F. A. C. et al. Epidermal growth factor receptors associated to cytoskeletal elements of the conformation and activity of EGFR dimers. epidermoid carcinoma (A431) cells. J. Cell Biol. 103, 87–94 (1986). Finally, monitoring the behaviour and reaction of signal-transduc- 15. Ishii, Y., Yoshida, T., Funatsu, T., Wazawa, T. & Yanagida, T. Fluorescence resonance energy transfer tion molecules using single-molecule techniques will be indispensable between single fluorophores attached to a coiled-coil protein in aqueous solution. Chem. Phys. 247, in fully understanding the mechanism of intracellular signalling. ᮀ 163–173 (1999). 16. Kinoshita, K., Itoh, H., Ishiwata, S., Nishizaka, T. & Hayakawa, T. Dual-view microscopy with a single camera: real-time imaging of molecular orientations and calcium. J. Cell Biol. 115, 67–73 (1991). 17. Campos-González, R. & Glenney, J. R. Jr Immunodetection of the ligand-activated receptor for Methods epidermal growth factor. Growth Factors 4, 305–316 (1991). Cell culture. 18. Pimplikar, S. W., Ikonen, E. & Simons, K. Basolateral protein transport in streptolysin O- A human carcinoma cell line, A431, was used 2 days after transfer of cells to a subculture. Before the permeabilized MDCK cells. J. Cell Biol. 125, 1025–1035 (1994). experiments, culture medium was exchanged to that without serum and phenol red overnight. 19. Lemmon, M. A. et al. Two EGF molecules contribute additively to stabilization of the EGFR dimer. EMBO J. 16, 281–294 (1997). Fluorescence labelling of proteins. 20. Hurwitz, D. R. et al. EGF induces increased ligand binding affinity and dimerization of soluble Recombinant mouse EGF (Higeta, Tokyo, Japan) was reacted with an N-hydroxysuccinidimyl ester of epidermal growth factor (EGF) receptor extracellular domain. J. Biol. Chem. 266, 22035–22043 Cy3 or Cy5 (Amersham) according to the manufacturer’s instructions. After a gel filtration using (1991). Sephadex G-15 (Pharmacia) to remove unreacted dye, Cy3 (or Cy5)-tagged EGF was further purified by 21. Sorokin, A., Lemmon, A., Ullrich, A. & Schlessinger, J. Stabilization of an active dimeric form of the reversed-phase column chromatograph26. A fraction with a dye:protein ratio of 0.9–1.1 was used. Mouse epidermal growth factor receptor by introduction of an inter-receptor disulfide bond. J. Biol. Chem. monoclonal anti-EGFR antibodies (EGFR1 (Genosys) and mAb74 (Transduction Laboratories)) were 269, 9752–9759 (1994). conjugated with Cy3 by the method used to label EGF. These antibodies can be labelled with multiple 22. Tzahar, E. et al. Bivalence of EGF-like ligands drives the ErbB signaling network. EMBO J. 16, 4938– Cy3 molecules. 4950 (1997). 23. Sako, Y. & Kusumi, A. Barriers for lateral diffusion of transferrin receptor in the plasma membrane TIR microscopy. as characterized by receptor dragging by laser tweezers: fence versus tether. J. Cell Biol. 129, 1559– We used an objective-type TIR fluorescence microscope3 based on an inverted microscope (IX-70, 1574 (1995). Olympus) equipped with an oil-immersion objective (PlanApo ×60, NA 1.4, Olympus). Specimens were 24. Kusumi, A. & Sako, Y. Cell surface organization by the membrane skeleton. Curr. Opin. Cell Biol. 8, illuminated with a SHG-Nd/YAG laser (model 142-0532-100, Lightwave, Mountain View, USA) with an 566–574 (1996). excitation power density of 10–100 µW µm–2. Images of the specimens were intensified by a microchannel 25. Sako, Y., Nagafuchi, A., Tsukita, S., Takeichi, M. & Kusumi, A. Cytoplasmic regulation of the plate (VS4-1845; Video Scope, Sterling, USA) and acquired by a chilled charge-coupled-device (CCD) movement of E-cadherin on the free cell surface as studied by optical tweezers and single particle camera (C5985; Hamamatsu, Hamamatsu, Japan) at video rate. tracking: corralling and tethering by the membrane skeleton. J. Cell Biol. 140, 1227–1240 (1998). 26. Carraway, K. L. III, Koland, J. G. & Cerione, R. A. Visualization of epidermal growth factor (EGF) Detection of EGFR phosphorylation. receptor aggregation in plasma membranes by fluorescence resonance energy transfer. Correlation of Cells were perforated with streptolysin O, and incubated with 20 U ml–1 calf thymus alkaline phosphatase receptor activation with aggregation. J. Biol. Chem. 264, 8699–8707 (1989). (Roche Molecular Biochemicals) in transport buffer (20 mM HEPES-KOH (pH 7.3), 110 mM ACKNOWLEDGEMENTS CH3COOK, 5 mM CH3COONa, 2 mM (CH3COO)2Mg, 1 mM EGTA, 1 mM dithiothreitol) for 30 min at 37 °C. After washing the cells twice with transport buffer, we stimulated the EGFR with 10 ng ml–1 Cy5– We thank T. Wazawa at the Single-Molecular Processes Project for the computer program with which to EGF in the presence of 1 mM ATP for 1 min at room temperature. The remaining Cy5–EGF in the measure fluorescence intensity; M. Murata for streptolysin O; and P. Conibear and F. Brozovich for medium was washed out, and cells were incubated with Cy3–mAb74 in the presence of 1 mM ATP in critical reading of the manuscript. transport buffer containing 1% bovine serum albumin.Cy3 and Cy5 were excited simultaneously by two Correspondence and requests for materials should be addressed to T.Y. lasers (532 and 632.8 nm). Fluorescence signals from Cy3 and Cy5 were separated by two dichroic Supplementary information is available on Nature Cell Biology’s World-Wide Web site (http:// mirrors (600 nm) and band-pass filters (565–595 nm for Cy3, and 650–690 nm for Cy5) in the dual-view cellbio.nature.com) or as paper copy from the London editorial office of Nature Cell Biology.